- Journal List
- Europe PMC Author Manuscripts
- PMC7616106
As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsem*nt of, or agreement with, the contents by NLM or the National Institutes of Health.
Learn more: PMC Disclaimer | PMC Copyright Notice
Eur Respir J. Author manuscript; available in PMC 2024 Jun 14.
Published in final edited form as:
Eur Respir J. 2024 Apr 1; 63(4): 2300127.
Published online 2024 Apr 4. doi:10.1183/13993003.00127-2023
PMCID: PMC7616106
EMSID: EMS196079
PMID: 37973176
An Zhao,1,2 Dr Eyjolfur Gudmundsson,1,2 Nesrin Mogulkoc, Prof,3 Dr Coline van Moorsel,4 Tamera J. Corte, Prof,5 Dr Pardeep Vasudev,1,2 Dr Chiara Romei,6 Dr Robert Chapman,7 Dr Tim J.M. Wallis,8 Dr Emma Denneny,7 Dr Tinne Goos,9,10 Dr Recep Savas,11 Dr Asia Ahmed,12 Dr Christopher J. Brereton,8 Dr Hendrik W. van Es,4 Dr Helen Jo,5 Dr Annalisa De Liperi,6 Dr Mark Duncan,12 Dr Katarina Pontoppidan,8 Dr Laurens J. De Sadeleer,10,13 Dr Frouke van Beek,4 Dr Joseph Barnett,14 Dr Gary Cross,15 Dr Alex Procter,12 Dr Marcel Veltkamp,4,16 Peter Hopkins, Prof,17 Yuben Moodley, Prof,18,19 Dr Alessandro Taliani,6 Dr Magali Taylor,12 Dr Stijn Verleden,20 Dr Laura Tavanti,21 Dr Marie Vermant,9,10 Dr Arjun Nair,12 Dr Iain Stewart,22 Sam M. Janes, Prof,23 Dr Alexandra L. Young,2,24 David Barber, Prof,25 Daniel C. Alexander, Prof,2 Joanna C. Porter, Prof,7 Athol U. Wells, Prof,26,27 Dr Mark G. Jones,8 Wim A. Wuyts, Prof,9,10 and Dr Joseph Jacob
1,2,23
Author information Copyright and License information PMC Disclaimer
The publisher's final edited version of this article is available at Eur Respir J
Associated Data
- Supplementary Materials
Abstract
Background
Idiopathic pulmonary fibrosis (IPF) with co-existent emphysema,termed combined pulmonary fibrosis and emphysema (CPFE) may associate withreduced forced vital capacity (FVC) declines compared to non-CPFE IPFpatients. We examined associations between mortality and functional measuresof disease progression in two IPF cohorts.
Methods
Visual emphysema presence (>0% emphysema) scored on computedtomography identified CPFE patients (CPFE:non-CPFE: derivationcohort=317:183; replication cohort=358:152), who were subgrouped using 10%,or 15% visual emphysema thresholds, and an unsupervised machine learningmodel considering emphysema and ILD extents. Baseline characteristics,1-year relative FVC and diffusing capacity of the lung for carbon monoxide(DLco) decline (linear mixed-effects models), and their associations withmortality (multivariable Cox regression models) were compared acrossnon-CPFE and CPFE subgroups.
Results
In both IPF cohorts, CPFE patients with ≥10% emphysema had agreater smoking history and lower baseline DLco compared to CPFE patientswith <10% emphysema. Using multivariable Cox regression analyses inpatients with ≥10% emphysema, 1-year DLco decline showed strongermortality associations than 1-year FVC decline. Results were maintained inpatients suitable for therapeutic IPF trials and in subjects subgrouped by≥15% emphysema and using unsupervised machine learning. Importantly,the unsupervised machine learning approach identified CPFE patients in whomFVC decline did not associate strongly with mortality. In non-CPFE IPFpatients, 1-year FVC declines ≥5% and ≥10% showed strongmortality associations.
Conclusion
When assessing disease progression in IPF, DLco decline should beconsidered in patients with ≥10% emphysema and a ≥5% 1-yearrelative FVC decline threshold considered in non-CPFE IPF patients.
Keywords: Combined pulmonary fibrosis and emphysema, mortality surrogates, idiopathic pulmonary fibrosis, computed tomography
Introduction
Emphysema is a common pulmonary finding on computed tomography (CT) imagingof idiopathic pulmonary fibrosis (IPF) patients [1]. The term combined pulmonary fibrosis and emphysema (CPFE) describesa potential clinical endotype characterized by the coexistence of upperlobe-predominant emphysema, lower lobe-predominant fibrosis and relativepreservation of forced vital capacity (FVC) in the context of a disproportionatelyreduced gas transfer (diffusing capacity of the lung for carbon monoxide, DLco)[1–3]. CPFE is highly heterogeneous in terms of the distribution andrelative extents of fibrosis and emphysema seen on CT.
CPFE patients are typically categorised using visual thresholds of emphysemaextent: >0%, ≥5%, ≥10%, ≥15%. It has been suggested thata subset of CPFE patients (≥15% emphysema) may manifest slower rates of FVCdecline than CPFE patients with lesser amounts of emphysema [4]. Despite the importance of fibrosis in driving FVC decline,fibrosis extent hasn’t been considered in prior definitions of CPFE [5]. Categorisation of CPFE patients using acombination of fibrosis and emphysema is possible using data-driven machine learningmethods. SuStaIn [6] is a machine learningmethod initially proposed for subtyping and modelling disease progression behaviourin dementia, which has been extended to COPD [7]. SuStaIn can identify disease subtypes with different progressionpatterns and can reconstruct their progression trajectories from cross-sectionaldata. A by-product of this approach would be the identification of patients indifferent CPFE subtypes who may benefit from different forms of disease progressionmonitoring, which in turn could inform clinical trial design.
In our study, we hypothesised that FVC decline, the most widely usedsurrogate for mortality prediction in IPF might show limited associations withmortality in independent CPFE populations with ≥10% and ≥15% emphysemascored visually on CT imaging, and in CPFE subgroups categorised by consideringrelative extents of interstitial lung disease (ILD) and emphysema. We furtherhypothesised that DLco decline could represent an alternative surrogate formortality in IPF patients with CPFE [5, 8].
Methods
Cohorts
Two independent IPF cohorts diagnosed by multidisciplinary teams werestudied. Patients with infection or cancer on baseline CT or who died within 3months of the baseline CT were excluded from the study. We studied two IPFcohorts so as to test whether DLco could be a consistent mortality surrogate inindependent IPF populations. The derivation cohort (n=500) derived from threecentres: Ege University Hospital, Izmir, Turkey; St Antonius Hospital,Nieuwegein, Netherlands; Pisa University Hospital, Italy. The replication cohort(n=510) derived from four centres: University Hospital Southampton NHSFoundation Trust, UK; University College London Hospitals NHS Foundation Trust,UK; University Hospitals Leuven, Belgium; Australian IPF registry, Australia.CONSORT diagrams for derivation cohort and replication cohort are shown in Supplementary Figure 1.Approval for this retrospective study of clinically indicated pulmonary functionand CT data was obtained from the local research ethics committees and LeedsEast Research Ethics Committee: 20/YH/0120.
Visual CT Scoring of Emphysema and ILD
Emphysema extent and fibrosis extent were visually scored in 6 lobes (thelingula was counted as the sixth lobe) by an experienced thoracic radiologist(JJ) with 16 year’s experience. Fibrosis extent comprised the sum ofground glass density (with overlying reticulation or traction bronchiectasis),reticulation, traction bronchiectasis and honeycomb cysts. Lobar extents ofemphysema/fibrosis were summed and divided by 6 to obtain a lung percentage ofemphysema/fibrosis.
For the purposes of this study, a patient was defined as having CPFE isthey had any emphysema on a CT. CPFE patients were subdivided in a primaryanalysis into those ≥10% emphysema (Figure1), and in a secondary analysis into those ≥15% emphysema. CTimaging in a random subset of 122 subjects was evaluated independently by tworadiologists (GC and JB: 3 and 4 years imaging experience respectively) toprovide an estimate of observer variation for semi-quantitative scores ofemphysema extent.
Figure 1
Computed tomography images of three subjects with 10% emphysema scoredvisually.
A 59-year-old male 5-pack-year ex-smoker with axial (a) and coronal (b) imagingshows extensive upper lobe paraseptal emphysema (black arrows) and alsocentrilobular emphysema (white arrows) in a predominantly upper lobedistribution. Fibrosis with traction bronchiectasis, ground glass opacificationand reticulation is seen in a lower zone predominant distribution. Figure c+dshow respectively axial and coronal images of mixed paraseptal (black arrows)and centrilobular emphysema (white arrows) in a 60-year-old male 17-pack-yearex-smoker. Axial images in a 72-year-old male 20-pack-year ex-smoker demonstratea predominantly paraseptal distribution of emphysema (black arrows) in the upper(e) and lower (f) lobes with minimal centrilobular emphysema (white arrow).
Statistical analysis
Data are presented as means and standard deviations unless otherwise stated.Two-sample t-tests were used for continuous variables, and chi-squared tests wereused for categorical variables. Kaplan-Meier survival plots and the log-rank testwere used to test for differences in survival between non-CPFE IPF patients, andCPFE patients in different subgroups (using emphysema thresholds or SuStaIn subtype)in both IPF cohorts. Subanalyses were performed for patients satisfying lungfunction criterion for inclusion into IPF therapeutic trials (percent predicted DLco>30%, percent predicted FVC >50%, and forced expiratory volume in thefirst second/FVC ratio >0.7).
FVC/DLco Decline Modelling
Linear mixed-effects (LME) models estimated absolute and relative 1-yearFVC decline and 1-year DLco decline. The trajectory of FVC for patients fromdifferent countries/centres was modelled separately by using the LME model.Fixed effects included: age at baseline CT date, sex, smoking history (never vs.ever), antifibrotics (never vs. ever), baseline percent predicted FVC (nearestto and within 3 months of baseline CT date), and time since baseline CT imagingdate. Each subject had a random intercept and random slope. FVC measurementsbetween baseline FVC date and 18 months after baseline CT date were used tobuild the LME model. Subjects were required to have had an FVC measurementwithin 3 months of the CT, and at least one further follow up FVC measurement tobe included in this analysis. Absolute and relative 1-year FVC declines werecalculated. For relative 1-year FVC decline, each follow-up FVC measurement(mls) was divided by baseline FVC (mls) and multiplied by 100 [9] and LME-predicted relative FVC percentagecalculated at 1 year. 1-year DLco decline was estimated using similar methods,with longitudinal DLco and baseline percent predicted DLco used in the LMEmodels. LME models were implemented with MATLAB (version R2019b, Mathworks,Natick, Massachusetts, US).
Machine Learning Delineation of CPFE Subtypes
Only patients with emphysema scored visually in any lobe were consideredfor SuStaIn CPFE analysis. Using baseline data alone, SuStaIn can identifydisease subtypes with distinct progression trajectories that describe theevolution of multiple biomarkers. The progression trajectory for an individualdisease subtype follows a linear z-score model, in which each biomarker ismodelled as a monotonically increasing piece-wise linear function [6, 7].Specifically, we used visually estimated fibrosis and emphysema extents withineach of the six lobes as biomarkers (12 biomarkers in total). The extent of eachbiomarker was divided by the interobserver variability (calculated using thesingle determination standard deviation) of the biomarker as scored by tworadiologists resulting in corresponding z-scores for the SuStaIn model. Thez-score indicates an abnormal level of a biomarker and the piece-wise lineartrajectory of each biomarker describes a continuous accumulation of abnormality:z-score = 0, 1, …, zmax. zmax is the maximumz-score a biomarker can reach at the end stage of a disease and this maximumscore can be a different number in different biomarkers. If we define thetransition of a biomarker from one z-score to the next z-score as a z-scoreevent, the trajectory of disease progression is a sequence of different z-scoreevents in the various biomarkers under consideration.
The process of fitting of the SuStaIn model aims to find the optimalnumber of subtypes of disease, the proportion of each subtype within thepopulation, and the order of z-score events for all biomarkers in each diseasesubtype. The trained SuStaIn model can then predict probabilities that anindividual belongs to a particular subtype and stage [6].
An underlying assumption of SuStaIn is that the biomarkers will show amonotonic increase. As emphysema develops slowly, and IPF patients have a shortsurvival time, it is less likely that an IPF patient without emphysema willdevelop emphysema during their lifetime. Accordingly, to avoid breaking theassumption that a biomarker will show a monotonic increase, only patients withemphysema scored visually in any lobe were considered for SuStaIn CPFEanalysis.
Cox Regression Modelling
In multivariable mixed-effects Cox regression models associations of FVCdecline and DLco decline with mortality were examined across IPF subtypes.Models were adjusted for age, sex, smoking history (never vs. ever),antifibrotic use (never vs. ever), and baseline disease severity (using percentpredicted DLco at baseline). Differences between different countries/centres ineach cohort were modelled by assigning a random intercept for each centre. Coxmodels were used with a minimum of 8 outcome events per predictor covariate[10]. Cox regression models weretested for proportional hazards assumption using the Schoenfeld residuals test.The Concordance index (C-index) compared the goodness of fit of Cox regressionmodels. P-values <0.01 were considered statistically significant. Allmixed-effects Cox regression analyses were implemented by R (version 4.0.3 withRstudio version 1.3.1093, Rstudio, Boston, Massachusetts, US).
Group Comparisons for FVC and DLco Decline
To investigate the impact of emphysema on FVC and DLco decline in thedifferent IPF subgroups (non-CPFE patients; CPFE patients classified usingemphysema thresholds or SuStaIn), proportions of patients with ≥5% and≥10% relative FVC decline in 1-year and ≥10% and ≥15%relative DLco decline in 1-year were calculated. Mean absolute 1-year FVCdecline (mls) and DLco decline (mls/min/mmHg) were also calculated for the threesubgroups. Analyses were performed in both IPF cohorts, with subanalyses insubjects fulfilling criteria for inclusion into IPF therapeutic trials.Chi-squared tests with Bonferroni-adjusted p-values were calculated forcategorical variables. A one-way ANOVA test examined differences in meanabsolute FVC decline (mls) with a post hoc Tukey Honest Significant Difference(HSD) test used to compare pairwise differences in subtypes.
Results
Baseline Characteristics
317/500 (63%) IPF patients in the derivation cohort had emphysema andwere defined as CPFE compared to 358/510 (7%) IPF patients with CPFE in thereplication cohort. CPFE patients were more likely to be smokers, had a higherpercent-predicted FVC and lower percent-predicted DLco than non-CPFEpatients.
Across the derivation and replication cohorts, CPFE patients with≥10% emphysema comprised greater numbers of smokers and had lowerbaseline percent predicted DLco compared to CPFE patients with <10%emphysema (Table 1). To power analyses,patients in both IPF cohorts fulfilling entry criteria for therapeutic trialswere combined into a single cohort (Supplementary Table 2). Baseline characteristics of CPFEpatients with emphysema above or below 15% in derivation and replication cohortsare shown in SupplementaryTable 3-4.
Table 1
Baseline characteristics of non-CPFE IPF patients and CPFE patients withemphysema below or above 10% in the derivation and replication cohorts.
| Cohort | Variable | Non-CPFE IPF patients | CPFE patients with emphy sema<10% | CPFE patients with emphysema>10% |
|---|---|---|---|---|
| Derivation cohort | Subjects (%) | 183 (36.6) | 174 (34.8) | 143 (28.6) |
| Age (years) | 67.8±9.2 | 66.9±9.1 | 65.0±9.1 | |
| Male (%) | 110/183 (60.1) | 143/174 (82.2) | 132/143 (92.3) | |
| Never-/ever-smokers (ever %) | 92/91 (49.7) | 38/133 (77.8) * | 8/134 (94.4) ** | |
| Visual fibrosis extent (%) | 38.7±14.6 | 36.3±14.1 | 40.8±13.5 | |
| Visual emphysema extent (%) | 0±0 | 4.8±2.3 | 20.4±8.8 | |
| FVC (% predicted, n) | 77.1±20.8 (158) | 80.11±20.2 (150) | 79.1±21.9 (122) | |
| DLco (% predicted, n) | 52.2±16.5 (151) | 51.6±15.1 (138) | 40.4±13.33 (116) | |
| Replication cohort | Subjects (%) | 152 (29.8) | 206 (40.4) | 152 (29.8) |
| Age (years) | 71.6±8.4 | 71.9±8.3 | 70.5±8.0 | |
| Male (%) | 96/152 (63.2) | 168/206 (81.6) | 128/152 (84.2) | |
| Never-/ever-smokers (ever %) | 78/74 (48.7) | 51/152 (74.9) † | 22/129 (85.4) †† | |
| Visual fibrosis extent (%) | 34.0±14.9 | 34.6±12.8 | 37.8±12.4 | |
| Visual emphysema extent (%) | 0±0 | 4.9±2.4 | 21.1±11.1 | |
| FVC (% predicted, n) | 84.5±21.1 (137) | 84.4±20.5 (184) | 86.6±18.9 (137) | |
| DLco (% predicted, n) | 55.2±15.1 (121) | 51.2±16.0 (176) | 40.7±11.2 (126) |
Open in a separate window
FVC: forced vital capacity; DLco: diffusing capacity of the lung forcarbon monoxide; CPFE: combined pulmonary fibrosis and emphysema; IPF:idiopathic pulmonary fibrosis; * 171 patients and ** 142 patients hadsmoking data available in derivation cohort;
* 171 patients and ** 142 patients had smoking data available inderivation cohort;
† 203 patients and †† 151patients had smoking data available in replication cohort.
The interobserver variation in visual emphysema scores for the subset of122 cases scored by two radiologists, measured using Cohens Kappa for 0%, 5%,10%, and 15% emphysema thresholds was: 0.2, 0.5, 0.61, 0.69, respectivelydemonstrating substantial agreement for a 10% visual emphysema threshold.
Machine Learning Model
Machine learning analyses of ILD and emphysema extents in the CPFEpopulation identified two distinct CPFE subtypes. One subtype(Fibrosis-Dominant CPFE; 60% of derivation cohort CPFEpatients and 61% of replication cohort CPFE patients) had much more extensivefibrosis at an early stage followed by a later emergence of emphysema (Figure 2). The second subtype(Matched-CPFE) demonstrated fibrosis and emphysemaworsening together, with later stages showing relatively more extensiveemphysema and less fibrosis compared to the Fibrosis-DominantCPFE subtype (Supplementary Table 5 and 6).
Figure 2
Identification of CPFE subtypes and subtype disease progression modelled bySuStaIn in the derivation cohort (a) and replication cohort (b). The rows showprogression patterns of fibrosis extent (in red) and emphysema extent (in blue)in 6 lung zones (upper, middle and lower) in the two CPFE subtypes identified bySuStaIn: Fibrosis-Dominant CPFE andMatched-CPFE. Seven disease stages are highlighted,expressed as z-score intervals. In the Fibrosis-Dominant CPFEsubtype comprising 60% of the derivation cohort and 60% of the replicationcohort (top two rows in (a) and (b)), fibrosis is more severe at an early stagefollowed by a later emergence of emphysema. In the Matched-CPFEsubtype comprising 40% of the derivation cohort and 39% of the replicationcohort (bottom two rows in (a) and (b)), fibrosis and emphysema get worsetogether, with later stages showing relatively more extensive emphysema and lessfibrosis compared to the Fibrosis-Dominant CPFE subtype. Theupper lobe predominance of emphysema seen at early disease stages no longerexists in the later stages of the Matched-CPFE subtype. CPFE:combined pulmonary fibrosis and emphysema. This figure was produced with theassistance of Servier Medical Art (https://smart.servier.com).
PFT Decline Analyses
Fewer CPFE patients with ≥10% emphysema reached the ≥10%or ≥5% 1-year FVC decline thresholds and had lower mean absolute FVCdeclines, though differences between groups did not reach statisticalsignificance (Table 2). Greater numbersof CPFE patients with ≥10% emphysema demonstrated 1-year DLco declines≥15%, though again results did not reach statistical significance (Table 3). Similar trends were found in thereplication cohort, patients fulfilling criteria to enter IPF therapeutic trials(Table 2 and and3),3), and when CPFE was categorized using a 15% emphysemathreshold or machine learning analyses (Supplementary Table 7 and 8).
Table 2
FVC decline analysis in different subgroups of IPF patients.
| Cohort | Subgroup | FVCdata available cases/all case | Relative 1-year FVCdecline (%) | Absolute 1-year FVCdecline (mls) | ||
|---|---|---|---|---|---|---|
| Numberof ≥10% (proportion) | Numberof ≥5% (proportion) | Mean | 95% CIof difference between subgroups | |||
| Derivation cohort | Non-CPFE | 150/183 | 51 (34%) | 81 (54%) | 163.50 | -117.78~84.55* |
| CPFE with emphysema <10% | 136/174 | 39 (28.68%) | 69 (50.74%) | 180.12 | -39.83~171.96# | |
| CPFE with emphysema ≥10% | 115/143 | 27 (23.48%) | 49(42.61%) | 97.43 | -190.92~25.55^ | |
| Replication cohort | Non-CPFE | 124/152 | 24 (19.35%) | 50 (40.32%) | 110.65 | -85.47~41.54* |
| CPFE with emphysema <10% | 170/206 | 37 (21.76%) | 75(44.12%) | 132.62 | -44.55~90.45# | |
| CPFE with emphysema ≥10% | 130/152 | 21 (16.15%) | 44 (33.85%) | 87.71 | -107.57~17.74^ | |
| Combined drugtrial cohort | Non-CPFE | 222/236 | 59 (26.58%) | 105 (47.30%) | 142.94 | -86.52~42.79* |
| CPFE with emphysema <10% | 240/261 | 57 (23.75%) | 113 (47.08%) | 164.81 | -42.64~104.13# | |
| CPFE with emphysema ≥10% | 150/157 | 29 (19.33%) | 56 (37.33%) | 112.19 | -124.88~19.65^ | |
Open in a separate window
The proportions of patients with more than 10% and 5% relative1-year FVC decline, and the mean of absolute 1-year FVC decline inderivation, replication cohorts and combined drug trial cohort (patientsfulfilling criteria to enter IPF therapeutic trials in derivation andreplication cohorts) are shown in this table. The number of subjects withavailable FVC decline versus the number of all subjects belonging to acertain subgroup is shown in n/n format. We also compared a) non-CPFE withCPFE with emphysema <1%, b) non-CPFE with CPFE with emphysema≥1%, c) CPFE with emphysema ≥10% and CPFE with emphysema<1%, in terms of the relative decline and absolute decline. We use *,# and ^ to denote comparison a), b), c) respectively in the table.None of the comparisons showed statistically significantdifferences. CPFE: combined pulmonary fibrosis and emphysema;IPF: idiopathic pulmonary fibrosis; FVC: forced vital capacity; CI:confidence interval.
Table 3
DLco decline analysis in different subgroups of IPF patients.
| Cohort | Subgroup | DLcodata available cases/all case | Relative 1-year DLcodecline (%) | Absolute 1-year DLcodecline (mls/min/mmHg) | ||
|---|---|---|---|---|---|---|
| Numberof ≥15% (proportion) | Number of ≥10% (proportion) | Mean | 95% CI of difference between subgroups | |||
| Derivation cohort | Non-CPFE | 132/183 | 52 (39.39%) | 73 (55.30%) | 645.39 | -881.03~129.87* |
| CPFE with emphysema <10% | 125/174 | 42 (33.60%) | 60 (48%) | 1020.97 | -752.33~301.34# | |
| CPFE with emphysema ≥10% | 107/143 | 42 (39.25%) | 59 (55.14%) | 870.88 | -683.49~383.31^ | |
| Replication cohort | Non-CPFE | 108/152 | 30 (27.78%) | 43 (39.81%) | 769.10 | -228.07~536.20* |
| CPFE with emphysema <10% | 161/206 | 38 (23.60%) | 67 (41.61%) | 615.04 | -222.08~597.87# | |
| CPFE with emphysema ≥10% | 117/152 | 42 (35.90%) | 64 (54.70%) | 581.21 | -407.07~339.41^ | |
| Combined drug trial cohort | Non-CPFE | 213/236 | 71 (33.33%) | 100 (46.95%) | 748.91 | -450.51~220.82* |
| CPFE with emphysema <10% | 238/261 | 66 (27.73%) | 112 (47.06%) | 863.75 | -448.18~316.55# | |
| CPFE with emphysema ≥10% | 146/157 | 54 (36.99%) | 80 (54.79%) | 814.72 | -423.13~325.08^ | |
Open in a separate window
The proportions of patients with more than 15% and 10% relative1-year DLco decline, and the mean of absolute 1-year DLco decline inderivation, replication cohorts and combined drug trial cohort (patientsfulfilling criteria to enter IPF therapeutic trials in derivation andreplication cohorts) are shown in this table. The number of subjects withavailable DLco decline versus the number of all subjects belonging to acertain subgroup is shown in n/n format. We also compared a) non-CPFE withCPFE with emphysema <10%, b) non-CPFE with CPFE with emphysema≥10%, c) CPFE with emphysema ≥10% and CPFE with emphysema<10%, in terms of the relative decline and absolute decline. We use*, # and ^ to denote comparison a), b), c) respectively in the table.None of the comparisons showed statistically significantdifferences. CPFE: combined pulmonary fibrosis and emphysema;IPF: idiopathic pulmonary fibrosis; DLco: diffusing capacity of the lung forcarbon monoxide; CI: confidence interval.
Survival Analyses
Kaplan-Meier survival plots (Figure3) demonstrated that in both cohorts, non-CPFE and CPFE patients with<10% emphysema had a significantly better prognosis than CPFE patientswith ≥10% emphysema. Results were maintained in patients fulfillingcriteria to enter IPF therapeutic trials and were similar when CPFE patientswere separated using a 15% emphysema threshold or machine learning analyses(Supplementary Figure 2and 3).
Figure 3
Kaplan-Meier curves of non-CPFE IPF patients (red), CPFE patients with emphysema<10% (green) and CPFE patients with emphysema ≥10% (blue) in thederivation cohort (a), the replication cohort (b), combined derivation andreplication cohort patients qualifying for therapeutic trials (c). Log-ranktests show a significant difference in mortality between the three subtypes inall three analyses.
Mortality Analysis for Visual Emphysema Thresholds
Multivariable Cox regression models adjusted for patient age, sex,smoking history (never vs. ever), antifibrotic use (never vs. ever), andbaseline percent predicted DLco showed that in non-CPFE patients, 5% and 10%1-year FVC decline thresholds showed strong associations with mortality inderivation (5% 1-year FVC decline: HR=3.82, 95% CI=2.10-6.95, p<0.0001;10% 1-year FVC decline: HR=4.26, 95% CI=2.42-7.50, p<0.0001) andreplication (5% 1-year FVC decline: HR=2.72, 95% CI=1.43-5.19, p=0.002; 10%1-year FVC decline: HR=2.73, 95% CI=1.37-5.44, p=0.004) cohorts (Table 4 and and5).5). Associations with mortality were maintained in patientsfulfilling criteria to enter IPF therapeutic trials (5% 1-year FVC decline:HR=3.27, 95% CI=2.03-5.25, p<0.0001; 10% 1-year FVC decline: HR=4.36, 95%CI=2.69-7.06, p<0.0001; Supplementary Table 9).
Table 4
Multivariable mixed-effects Cox proportional hazards regression models innon-CPFE patients and the two CPFE subgroups in the derivation IPFcohort.
| Subgroup | Baseline severity and PFTschanges models | C-index | p-value | Hazard ratio | 95% CI | |
|---|---|---|---|---|---|---|
| Lower | Upper | |||||
| Non-CPFE IPFpatients (n=130, 61 deaths) | 1-year FVC relative decline | 0.821 | 3.02×10-8 | 1.082 | 1.052 | 1.113 |
| Binary 1-year FVC decline (5%) | 0.805 | 1.09×10-5 | 3.824 | 2.104 | 6.953 | |
| Binary 1-year FVC decline (10%) | 0.811 | 4.96×10-7 | 4.261 | 2.422 | 7.497 | |
| 1-year DLco relative decline | 0.803 | 0.0001 | 1.038 | 1.018 | 1.058 | |
| Binary 1-year DLco decline (10%) | 0.800 | 0.0010 | 2.764 | 1.511 | 5.055 | |
| Binary 1-year DLco decline (15%) | 0.811 | 4.69×10-7 | 4.211 | 2.407 | 7.366 | |
| CPFEpatients with emphysema < 10%(n=119, 63 deaths) | 1-year FVC relative decline | 0.716 | 6.46×10-5 | 1.051 | 1.026 | 1.077 |
| Binary 1-year FVC decline (5%) | 0.721 | 0.0001 | 3.000 | 1.705 | 5.279 | |
| Binary 1-year FVC decline (10%) | 0.685 | 0.025 | 1.983 | 1.091 | 3.604 | |
| 1 -year DLco relative decline | 0.727 | 0.0003 | 1.035 | 1.016 | 1.055 | |
| Binary 1-year DLco decline (10%) | 0.682 | 0.173 | 1.453 | 0.849 | 2.486 | |
| Binary 1-year DLco decline (15%) | 0.696 | 0.017 | 1.979 | 1.131 | 3.464 | |
| CPFEpatients with emphysema ≥1% (n=103,73 deaths) | 1-year FVC relative decline | 0.714 | 0.008 | 1.034 | 1.009 | 1.061 |
| Binary 1-year FVC decline (5%) | 0.714 | 0.016 | 1.868 | 1.126 | 3.100 | |
| Binary 1-year FVC decline (10%) | 0.715 | 0.002 | 2.540 | 1.421 | 4.539 | |
| 1-year DLco relative decline | 0.732 | 1.24×10-5 | 1.033 | 1.018 | 1.049 | |
| Binary 1-year DLco decline (1%) | 0.703 | 0.058 | 1.619 | 0.983 | 2.665 | |
| Binary 1-year DLco decline (15%) | 0.732 | 7.61×10-5 | 2.674 | 1.643 | 4.353 | |
Open in a separate window
Multivariable mixed-effects Cox regression models were used toinvestigate associations with mortality for 1-year FVC decline and 1-yearDLco decline after adjusting for patient age, sex, smoking status (neverversus ever), antifibrotic use (never versus ever) and baseline diseaseseverity estimated using DLco. Binary 1-year FVC decline uses 5% and 10%relative decline as thresholds, and binary 1-year DLco decline uses 10% and15% relative decline as thresholds. Separate centres/countries within thederivation cohort were modelled as multilevel with random effects betweencentres/countries (a random intercept per centre/country). All models passedSchoenfeld residuals test. CPFE: combined pulmonary fibrosis and emphysema;IPF: idiopathic pulmonary fibrosis; PFT: pulmonary function test; FVC:forced vital capacity; DLco: diffusing capacity of the lung for carbonmonoxide; C-index: concordance index; CI: confidence interval.
Table 5
Multivariable mixed-effects Cox proportional hazards regression models innon-CPFE patients and the two CPFE subgroups in the replication IPFcohort.
| Subgroup | Baseline severity and PFTschanges models | C-index | p-value | Hazard ratio | 95% CI | |
|---|---|---|---|---|---|---|
| Lower | Upper | |||||
| Non-CPFE IPFpatients (n=108, 45 deaths) | 1-year FVC relative decline | 0.823 | 8.65×10-5 | 1.086 | 1.042 | 1.132 |
| Binary 1-year FVC decline (5%) | 0.827 | 0.002 | 2.719 | 1.425 | 5.187 | |
| Binary 1-year FVC decline (10%) | 0.817 | 0.004 | 2.733 | 1.374 | 5.437 | |
| 1 -year DLco relative decline | 0.822 | 0.019 | 1.032 | 1.005 | 1.059 | |
| Binary 1-year DLco decline (10%) | 0.835 | 0.013 | 2.373 | 1.201 | 4.688 | |
| Binary 1-year DLco decline (15%) | 0.835 | 0.006 | 2.693 | 1.336 | 5.428 | |
| CPFEpatients with emphysema < 10%(n=159, 83 deaths) | 1-year FVC relative decline | 0.754 | 0.001 | 1.055 | 1.022 | 1.089 |
| Binary 1-year FVC decline (5%) | 0.763 | 0.004 | 1.960 | 1.246 | 3.083 | |
| Binary 1-year FVC decline (10%) | 0.767 | 9.27×10-5 | 2.704 | 1.642 | 4.453 | |
| 1 -year DLco relative decline | 0.776 | 2.87×10-5 | 1.032 | 1.017 | 1.047 | |
| Binary 1-year DLco decline (10%) | 0.772 | 0.0005 | 2.252 | 1.424 | 3.561 | |
| Binary 1-year DLco decline (15%) | 0.768 | 0.0001 | 2.781 | 1.659 | 4.661 | |
| CPFEpatients with emphysema ≥1% (n=115, 70deaths) | 1-year FVC relative decline | 0.705 | 0.130 | 1.024 | 0.993 | 1.056 |
| Binary 1-year FVC decline (5%) | 0.689 | 0.707 | 1.105 | 0.656 | 1.863 | |
| Binary 1-year FVC decline (10%) | 0.706 | 0.035 | 2.028 | 1.053 | 3.906 | |
| 1 -year DLco relative decline | 0.720 | 0.001 | 1.030 | 1.012 | 1.049 | |
| Binary 1-year DLco decline (10%) | 0.716 | 0.0004 | 2.672 | 1.546 | 4.617 | |
| Binary 1-year DLco decline (15%) | 0.729 | 1.04×10-5 | 3.883 | 2.124 | 7.097 | |
Open in a separate window
Multivariable mixed-effects Cox regression models were used toinvestigate associations with mortality for 1-year FVC decline and 1-yearDLco decline after adjusting for patient age, sex, smoking status (neverversus ever), antifibrotic use (never versus ever) and baseline diseaseseverity estimated using DLco. Binary 1-year FVC decline uses 5% and 10%relative decline as thresholds, and binary 1-year DLco decline uses 10% and15% relative decline as thresholds. Separate centres/countries within thereplication cohort were modelled as multilevel with random effects betweencentres/countries (a random intercept per centre/country). All models passedSchoenfeld residuals test. CPFE: combined pulmonary fibrosis and emphysema;IPF: idiopathic pulmonary fibrosis; PFT: pulmonary function test; FVC:forced vital capacity; DLco: diffusing capacity of the lung for carbonmonoxide; C-index: concordance index; CI: confidence interval.
For CPFE patients with ≥10% emphysema (derivation cohortn=103/352 (29%); replication cohort n=115/382 (30%)), in multivariable analyses,1-year relative DLco decline showed a stronger association with mortality than1-year relative FVC decline in derivation (DLco decline: HR=1.03, 95%CI=1.02-1.05, p<0.0001; FVC decline: HR=1.03, 95% CI=1.01-1.06, p=0.008)and replication (DLco decline: HR=1.03, 95% CI=1.01-1.05, p=0.001; FVC decline:HR=1.02, 95% CI=0.99-1.06, p=0.13) cohorts (Table 4 and and5).5). When DLcothresholds were examined in CPFE patients with ≥10% emphysema,≥15% 1-year relative DLco decline showed stronger associations withmortality than ≥10% 1-year relative FVC decline in derivation(≥15% 1-year DLco decline: HR=2.67, 95% CI=1.64-4.35, p<0.0001;≥10% 1-year FVC decline: HR=2.54, 95% CI=1.42-4.54, p=0.002) andreplication (≥15% 1-year DLco decline: HR=3.88, 95% CI=2.12-7.10,p<0.0001; ≥10% 1-year FVC decline: HR=2.03, 95% CI=1.05-3.91,p=0.04) cohorts. In subjects eligible for inclusion into IPF therapeutic trials(where 144/589 (24%) patients had ≥10% emphysema) 1-year relative DLcodecline (HR=1.04, 95% CI=1.03-1.06, p<0.0001) showed strongerassociations with mortality than 1-year relative FVC decline (HR=1.05, 95%CI=1.02-1.08, p=0.0006) on multivariable Cox regression analyses (Supplementary Table 9).Similar trends were observed in multivariable analyses performed in CPFEpatients with ≥15% emphysema (Supplementary Table 10-12).
Mortality Analyses of Machine Learning Derived CPFE Subgroups
Trends seen for the 10% visual emphysema threshold were again replicatedwhen CPFE patients were separated using machine learning analyses thatconsidered ILD and emphysema extents. The Matched-CPFE cohortalso delineated patients in whom FVC decline proved a poor surrogate formortality. Importantly, in the Matched-CPFE cohort, DLcodecline, whether measured as relative decline in percent-predicted DLco(derivation: HR=1.04, 95% CI=1.02-1.05, p<0.0001; replication: HR=1.03,95% CI=1.01-1.05, p=0.001, clinical trial cohort: HR=1.04, 95% CI=1.03-1.06,p<0.0001) or a ≥15% DLco threshold (derivation: HR=2.63, 95%CI=1.54-4.52, p=0.0004; replication: HR=4.86, 95% CI=2.39-9.90, p<0.0001,clinical trial cohort: HR=3.61, 95% CI=2.16-6.02, p<0.0001) remained astrong surrogate for mortality (Supplementary Table 13-15). This was less evident for FVCdecline (measured in mls) whether expressed as a continuous relative declinepercentage (derivation: HR=1.04, 95% CI=1.01-1.07, p=0.006; replication:HR=1.02, 95% CI=0.99-1.06, p=0.23, clinical trial cohort: HR=1.06, 95%CI=1.03-1.09, p=0.0006) or a ≥10% FVC decline threshold (derivation:HR=2.48, 95% CI=1.22-5.07, p=0.01; replication: HR=2.36, 95% CI=1.14-4.91,p=0.02, clinical trial cohort: HR=2.67, 95% CI=1.42-5.02, p=0.002).
Discussion
Our study evaluated functional indicators of disease progression in IPFpatients with emphysema that have been the key mortality surrogates used in clinicalcare and therapeutic trials. We identified three important findings across two IPFpopulations: Firstly, we demonstrated the limited associations between relative FVCdecline and mortality in CPFE patients with ≥10% and ≥15% emphysema,and conversely the strong associations with mortality for relative DLco decline inthe same subgroups. Second, our machine learning model identified a subgroup of CPFEpatients where a relatively greater amount of emphysema compared to ILD accentuatedthe limited associations between ILD-driven FVC decline and mortality in these CPFEpatients. Lastly, in non-CPFE patients we showed that FVC decline is a powerfulmeasure of IPF progression showing strong associations with mortality at both≥5% and ≥10% 1-year FVC decline thresholds.
FVC decline occupies a cardinal role in the assessment of diseaseprogression in IPF as it has been shown to be a strong surrogate for mortality[11]. The demonstration however that FVCdecline may be curtailed in IPF patients with ≥15% [4] emphysema raised the question of whether FVC decline remaineda surrogate for mortality in IPF patients with more extensive emphysema. Only oneother study, by Schmidt et al [8], which wasrelatively underpowered (n=42) for subjects with moderate/severe emphysema (definedas emphysema at least as extensive as ILD), addressed this question and found thatFVC decline did not associate with mortality at 12 months. Other studies consideringIPF patients regardless of emphysema presence/extent have shown strong associationsbetween mortality and other functional decline measures/thresholds including: DLcodecline thresholds of ≥10% [12] and15% [13], and FVC declines of ≥5%[14–16].
An explanation for the poor association between FVC decline and mortality inpatients with more extensive emphysema may relate to the impact of fibrosis whenencroaching on areas of emphysema. Emphysematous regions of lung commonlydemonstrate air trapping as thickened small airways collapse on expiration. Fibroticprocesses however can irreversibly pull open small airways. The supervening tractionbronchiolectasis can result in emphysematous airspaces being ventilated, therebyartificially preserving FVC. In IPF patients with emphysema, as fibrosis progressesand extends to involve the upper zones of the lungs, more emphysematous lung maybecome incorporated into the expiratory lung volume over time. A consequence may begreater heterogeneity in expiratory lung volumes, superimposing considerable noiseto an overarching pattern of progressive FVC decline. This effect is likely to bemore pronounced in patients with more extensive emphysema.
One limitation in prior definitions of CPFE has been the focus on emphysemaextent alone as the sole arbiter for categorising a CPFE endotype. A recentATS/ERS/ALAT/JRS research statement identified a 5% emphysema threshold as aresearch definition for CPFE patients, whilst suggesting a 15% emphysema thresholdfor classifying a CPFE clinical syndrome [5].In our study we found that a 10% emphysema threshold (which showed substantial CTobserver agreement) may represent a better cut-off than a 15% emphysema threshold toidentify a CPFE population disenfranchised by the use of FVC as a sole measure ofdisease progression.
A further challenge with CPFE definitions being determined by emphysemathresholds is that FVC decline is primarily driven by ILD progression rather thanemphysema progression. Our unsupervised machine learning model (SuStaIn) consideredboth fibrosis and emphysema when subtyping patients and replicated the strongassociation of DLco decline and mortality in patients with more extensive emphysemaseen in CPFE patients with ≥10% emphysema. By considering ILD extent inrelation to emphysema extent, the SuStaIn model delineated of a subgroup of CPFEpatients, fulfilling criteria to enter IPF therapeutic trials, where FVC decline didnot associate strongly with mortality.
Prior studies have shown associations between DLco decline and mortality inIPF [8, 12, 13, 17–19] but havenot analysed the impact of emphysema on DLco trends. DLco decline has generally beenless consistent in its links with mortality than FVC decline in IPF patients [20]. Yet DLco decline may have particularrelevance in subsets of IPF patients [21].For example, the strong mortality signal for DLco decline seen in CPFE patients withmore extensive emphysema could reflect progressive localised pulmonary hypertensioncomplicating CPFE patients with more extensive emphysema [22, 23]. Our studyfindings suggest that in IPF patients with extensive emphysema a composite endpointof FVC decline ≥10% or DLco decline ≥15% should be considered whenassessing disease progression.
There were limitations to the current study. A single observer scored theCTs for fibrosis and emphysema. For studies to be clinically meaningful, they haveto be suitably powered, and this requires the careful evaluation of large IPFpopulations. This is challenging with a current limited availability of radiologistsand would occur more commonly in specialist ILD centres. The single read of CTs inthis study aligns with other large scale IPF studies where pragmatic considerationsrequired assessment of CTs by a single specialist [24, 25]. Similar functionalmeasures and IPF subgroups proportions across both study cohorts provide reassurancefor the validity of the visual CT scores. The improvement in observer agreement athigher emphysema thresholds (even amongst less experienced radiologists) addsconfidence to the reliability of visual scores at an emphysema threshold of 1%. Thisalso aligns with prior work [26]demonstrating improved interobserver agreement at emphysema extent categories of 10%and 15% versus 0% and 5%. The computer algorithm SuStaIn is not routinely availableto clinicians at present, but was used to show the impact of considering ILD extentin the classification of CPFE subtypes. There was also missing data for longitudinalPFTs, reducing the sample size of both cohorts in the analyses of lung functiondecline. No imputation was performed however as we wanted the analyses to reflectthe recorded functional status of the patients. Lastly, whilst we would have likedto have fully automated our machine learning model, using computationally quantifiedemphysema as an objective measure of disease, no existing automated tools canreliably distinguish emphysema from honeycombing and traction bronchiectasis.
In conclusion, annual relative DLco decline was shown to be a bettermortality surrogate for patients with more than 10% emphysema than relative FVCdecline. Findings were validated by a data-driven machine learning method thatconsiders emphysema and ILD extents when defining patients with more extensiveemphysema. These observations may be useful in clinical trial design to identifysubjects where FVC decline is a poor disease progression measure. A 5% 1-yearrelative FVC decline threshold however was found to be a strong mortality indicatorin non-CPFE IPF patients.
Supplementary Material
Supplementary Materials
Click here to view.(981K, pdf)
Acknowledgments
This research was funded in whole or in part by the Wellcome Trust [209553/Z/17/Z].For the purpose of open access, the author has applied a CC-BY public copyrightlicence to any author accepted manuscript version arising from this submission. Thisproject, JJ, EG, ED, SMJ and JCP were also supported by the NIHR UCLH BiomedicalResearch Centre, UK. MGJ, TJMW and CJB acknowledge the support of the NIHRSouthampton Biomedical Research Centre. AZ was supported by CSC-UCL Joint ResearchScholarship. The Australian IPF Registry is an initiative of Lung FoundationAustralia and is supported by Foundation partners Boehringer Ingelheim, RocheProducts Pty. Limited.
Footnotes
Disclosure of Conflicts of Interest
JJ reports fees from Boehringer Ingelheim, Roche, NHSX, Takeda andGlaxoSmithKline unrelated to the submitted work. JJ was supported by WellcomeTrust Clinical Research Career Development Fellowship 209553/Z/17/Z and the NIHRBiomedical Research Centre at University College London. NM reports grantTUBITAK (EJP Rare Disease project “COCOS-IPF”), fees fromBoehringer, Ingelheim, Roche, and Nobel Turkey unrelated to the submitted work.NM received support for travel to ATS 2020 and ATS 2021 from Roche, and to ERS2020 from Actelion. TJC reports unrestricted educational grants from BoehringerIngelheim, Roche, Biogen, and Galapagos. TJC reports fees from Roche, BMS,Boehringer Ingelheim, Vicore, DevPro. TJC received assistance for travel tomeetings from Boehringer Ingelheim. TJC reports participation on a Data SafetyMonitoring Board or Advisory Board of Roche, BMS, Boehringer Ingelheim, Vicore,Ad Alta, Bridge Biotherapeutics, DevPro. PV reports financial interests fromBlackford Analysis. TG is supported by Research Foundation-Flanders(FWO)-1S73921N. LJDS is supported by Marie Sklodowska-Curie actions postdoctoralfellowship within the European Union’s Horizon Europe research andinnovation programme. HJ reports fees from Boehringer Ingelheim and Roche. HJreceived assistance for travel to meetings from Boehringer Ingelheim and Roche.SV reports consultancy fees from Boehringer-Ingelheim and Sanofi. MV issupported by FWO (Research Flanders Foundation) Fellowship. SMJ reports feesfrom Astra-Zeneca, Bard1 Bioscience, Achilles Therapeutics, and Jansen unrelatedto the submitted work. SMJ received assistance for travel to meetings from AstraZeneca to American Thoracic Conference 2018 and from Takeda to World ConferenceLung Cancer 2019 and is the Investigator Lead on grants from GRAIL Inc,GlaxoSmithKline plc and Owlstone. AUW reports personal fees and non-financialsupport from Boehringer Ingelheim, Bayer and Roche Pharmaceuticals; and personalfees from Blade, outside of the submitted work. AZ, EG, CvM, CR, RC, TJMW, ED,RS, AA, CJB, HWvE, ADL, MD, KP, FvB, JB, GC, AP, MV, PH, YM, AT, MT, LT, AN, IS,ALY, DB, DCA, JCP, MGJ, WAW report no relevant conflicts of interest.
Contributed by
Author Contributions
AZ, EG, IS, ALY, DB, DCA, AUW, and JJ contributed to study design anddata interpretation. AZ, EG, NM, MGJ, CvM, TJC, PV, CR, RC, TJMW, ED, TG, RS,AA, CJB, HWvE, HJ, ADL, MD, KP, LJDS, FvB, JB, GC, AP, MV, PH, YM, AT, MT, SV,LT, MV, AN, SMJ, JCP, MGJ, WAW and JJ were responsible for data acquisition. AZ,EG, IS, and JJ contributed to the statistical analysis. AZ and JJ prepared thefirst draft of the manuscript. AZ and JJ were responsible for study dataintegrity. All authors reviewed the manuscript and approved the final submittedversion.
References
1. King CS, Nathan SD. Idiopathic pulmonary fibrosis: effects and optimal management ofcomorbidities. Lancet Respir Med. 2017;5:72–84. [PubMed] [Google Scholar]
2. Lin H, Jiang S. Combined pulmonary fibrosis and emphysema (CPFE): an entitydifferent from emphysema or pulmonary fibrosis alone. J Thorac Dis. 2015;7:767–779. [PMC free article] [PubMed] [Google Scholar]
3. Cottin V, Nunes H, Brillet P, et al. Combined pulmonary fibrosis and emphysema: a distinctunderrecognised entity. Eur Respir J. 2005;26:586–593. [PubMed] [Google Scholar]
4. Cottin V, Hansell DM, Sverzellati N, et al. Effect of Emphysema Extent on Serial Lung Function in Patientswith Idiopathic Pulmonary Fibrosis. Am J Respir Crit Care Med. 2017;196:1162–1171. [PubMed] [Google Scholar]
5. Cottin V, Selman M, Inoue Y, et al. Syndrome of Combined Pulmonary Fibrosis and Emphysema: AnOfficial ATS/ERS/JRS/ALAT Research Statement. Am J Respir Crit Care Med. 2022;206:e7–e41. [PMC free article] [PubMed] [Google Scholar]
6. Young AL, Marinescu RV, Oxtoby NP, et al. Uncovering the heterogeneity and temporal complexity ofneurodegenerative diseases with Subtype and Stage Inference. Nat Commun. 2018;9:4273. [PMC free article] [PubMed] [Google Scholar]
7. Young AL, Bragman FJS, Rangelov B, et al. Disease Progression Modeling in Chronic Obstructive PulmonaryDisease. Am J Respir Crit Care Med. 2020;201:294–302. [PMC free article] [PubMed] [Google Scholar]
8. Schmidt SL, Nambiar AM, Tayob N, et al. Pulmonary function measures predict mortality differently in IPFversus combined pulmonary fibrosis and emphysema. Eur Respir J. 2011;38:176–183. [PMC free article] [PubMed] [Google Scholar]
9. Jacob J, Bartholmai BJ, Rajagopalan S, et al. Predicting Outcomes in Idiopathic Pulmonary Fibrosis UsingAutomated Computed Tomographic Analysis. Am J Respir Crit Care Med. 2018;198:767–776. [PMC free article] [PubMed] [Google Scholar]
10. Vittinghoff E, McCulloch CE. Relaxing the Rule of Ten Events per Variable in Logistic and CoxRegression. Am J Epidemiol. 2007;165:710–718. [PubMed] [Google Scholar]
11. Molina Molina M, Hart E, Lesher B, et al. Association between FVC and mortality or survival in idiopathicpulmonary fibrosis: a systematic literature review. Eur Respir J. 2021;58:PA3753 [Google Scholar]
12. Salisbury ML, Xia M, Zhou Y, et al. Idiopathic Pulmonary Fibrosis: Gender-Age-Physiology Index Stagefor Predicting Future Lung Function Decline. Chest. 2016;149:491–498. [PMC free article] [PubMed] [Google Scholar]
13. Doubková M, Švancara J, Svoboda M, et al. EMPIRE Registry, Czech Part: Impact of demographics, pulmonaryfunction and HRCT on survival and clinical course in idiopathic pulmonaryfibrosis. Clin Respir J. 2018;12:1526–1535. [PubMed] [Google Scholar]
14. du Bois RM, Weycker D, Albera C, et al. Ascertainment of Individual Risk of Mortality for Patients withIdiopathic Pulmonary Fibrosis. Am J Respir Crit Care Med. 2011;184:459–466. [PubMed] [Google Scholar]
15. du Bois RM, Albera C, Bradford WZ, et al. 6-minute walk distance is an independent predictor of mortalityin patients with idiopathic pulmonary fibrosis. Eur Respir J. 2014;43:1421 LP–1429. [PubMed] [Google Scholar]
16. Reichmann WM, Yu YF, Macaulay D, et al. Change in forced vital capacity and associated subsequentoutcomes in patients with newly diagnosed idiopathic pulmonaryfibrosis. BMC Pulm Med. 2015;15:167. [PMC free article] [PubMed] [Google Scholar]
17. Bodlet A, Maury G, Jamart J, et al. Influence of radiological emphysema on lung function test inidiopathic pulmonary fibrosis. Respir Med. 2013;107:1781–1788. [PubMed] [Google Scholar]
18. Gonzalez Taylor, Maher T. Predicting mortality in idiopathic pulmonary fibrosis. Whichparameters should be used to determine eligibility for treatment? Analysisof a UK prospective cohort. Eur Respir J. 2016;48:OA282 [Google Scholar]
19. Zurkova M, Kriegova E, Kolek V, et al. Effect of pirfenidone on lung function decline and survival: 5-yrexperience from a real-life IPF cohort from the Czech EMPIREregistry. Respir Res. 2019;20:16. [PMC free article] [PubMed] [Google Scholar]
20. Raghu G, Remy-Jardin M, Richeldi L, et al. Idiopathic Pulmonary Fibrosis (an Update) and ProgressivePulmonary Fibrosis in Adults: An Official ATS/ERS/JRS/ALAT Clinical PracticeGuideline. Am J Respir Crit Care Med. 2022;205:e18–e47. [PMC free article] [PubMed] [Google Scholar]
21. Akagi T, Matsumoto T, Harada T, et al. Coexistent emphysema delays the decrease of vital capacity inidiopathic pulmonary fibrosis. Respir Med. 2009;103:1209–1215. [PubMed] [Google Scholar]
22. Mejía M, Carrillo G, Rojas-Serrano J, et al. Idiopathic Pulmonary Fibrosis and Emphysema: Decreased SurvivalAssociated With Severe Pulmonary Arterial Hypertension. Chest. 2009;136:10–15. [PubMed] [Google Scholar]
23. Cottin V, Le Pavec J, Prévot G, et al. Pulmonary hypertension in patients with combined pulmonaryfibrosis and emphysema syndrome. Eur Respir J. 2010;35:105 LP–111. [PubMed] [Google Scholar]
24. Walsh SLFF, Calandriello L, Silva M, et al. Deep learning for classifying fibrotic lung disease onhigh-resolution computed tomography: a case-cohort study. Lancet Respir Med. 2018;6:837–845. [PubMed] [Google Scholar]
25. Salisbury ML, Hewlett JC, Ding G, et al. Development and Progression of Radiologic Abnormalities inIndividuals at Risk for Familial Interstitial Lung Disease. Am J Respir Crit Care Med. 2020;201:1230–1239. [PMC free article] [PubMed] [Google Scholar]
26. Jacob J, Odink A, Brun AL, et al. Functional associations of pleuroparenchymal fibroelastosis andemphysema with hypersensitivity pneumonitis. Respir Med. 2018;138:95–101. [PMC free article] [PubMed] [Google Scholar]
